Endotracheal tube mucus as a source of airway mucus for rheological study

X Matthew R. Markovetz,1 Durai B. Subramani,1 William J. Kissner,1 Cameron B. Morrison,1

Ian C. Garbarine,1 Andrew Ghio,2 Kathryn A. Ramsey,1 Harendra Arora,3,4 Priya Kumar,3,4

David B. Nix,5 Tadahiro Kumagai,5 Thomas M. Krunkosky,6 Duncan C. Krause,6 Giorgia Radicioni,1

Neil E. Alexis,7 Mehmet Kesimer,1,8 Michael Tiemeyer,5 Richard C. Boucher,1 Camille Ehre,1 andDavid B. Hill1,9

1Marsico Lung Institute, University of North Carolina, Chapel Hill, North Carolina; 2National Health and EnvironmentalEffects Research Laboratory, United States Environmental Protection Agency, Research Triangle Park, North Carolina;3Department of Anesthesiology, University of North Carolina, Chapel Hill, North Carolina; 4Outcomes Research Consortium,Cleveland, Ohio; 5Complex Carbohydrate Research Center, University of Georgia, Athens, Georgia; 6Department ofMicrobiology, University of Georgia, Athens, Georgia; 7Center for Environmental Medicine, Asthma, and Lung Biology,University of North Carolina, Chapel Hill, North Carolina; 8Department of Pathology and Laboratory Medicine, Universityof North Carolina, Chapel Hill, North Carolina; and 9Department of Physics and Astronomy, University of North Carolina,Chapel Hill, North Carolina

Submitted 17 June 2019; accepted in final form 1 August 2019

Markovetz MR, Subramani DB, Kissner WJ, Morrison CB,Garbarine IC, Ghio A, Ramsey KA, Arora H, Kumar P, Nix DB,Kumagai T, Krunkosky TM, Krause DC, Radicioni G, Alexis NE,Kesimer M, Tiemeyer M, Boucher RC, Ehre C, Hill DB. Endotra-cheal tube mucus as a source of airway mucus for rheological study.Am J Physiol Lung Cell Mol Physiol 317: L498–L509, 2019. Firstpublished August 7, 2019; doi:10.1152/ajplung.00238.2019.—Muco-obstructive lung diseases (MOLDs), like cystic fibrosis and chronicobstructive pulmonary disease, affect a spectrum of subjects globally.In MOLDs, the airway mucus becomes hyperconcentrated, increasingosmotic and viscoelastic moduli and impairing mucus clearance.MOLD research requires relevant sources of healthy airway mucus forexperimental manipulation and analysis. Mucus collected from endo-tracheal tubes (ETT) may represent such a source with benefits, e.g.,in vivo production, over canonical sample types such as sputum orhuman bronchial epithelial (HBE) mucus. Ionic and biochemicalcompositions of ETT mucus from healthy human subjects werecharacterized and a stock of pooled ETT samples generated. PooledETT mucus exhibited concentration-dependent rheologic propertiesthat agreed across spatial scales with reported individual ETT samplesand HBE mucus. We suggest that the practical benefits compared withother sample types make ETT mucus potentially useful for MOLDresearch.

cystic fibrosis; muco-obstructive lung disease; mucus; mucus bio-chemistry; mucus biophysics

INTRODUCTION

The airway surface liquid (ASL) is a protective bilayercomposed of the mucus layer and the periciliary layer (PCL)(7). MUC5B and MUC5AC are the primary secreted mucinsin airway mucus and, with �1,000 other proteins, providethe mucus layer with its characteristic viscoelastic proper-ties (23, 45). The PCL consists of grafted (tethered) poly-meric mucins and other glycoproteins that are bound to and

lubricate the cilia as they coordinately beat to propel themucus from the lung (7, 8).

In health, the mucus and periciliary layer are coordinatelyhydrated to maintain efficient mucus clearance (7). Ciliary beatin health in a well-hydrated PCL environment is unrestrictedby friction or negative osmotic interactions with the mucuslayer. However, in muco-obstructive lung diseases (MOLDs)such as chronic obstructive pulmonary disease and cysticfibrosis (CF), the mucus layer becomes hyperconcentrated (i.e.,isotonically dehydrated) (16, 19, 22, 32). The hyperconcen-trated mucus (i.e., �4% solids) has increased osmotic andviscoelastic moduli that slow clearance and produce mucusstasis (2, 6). Static mucus is proinflammatory and the site ofbacterial infection, which generates a positive-feedback mech-anism that further increases mucus secretion/concentration inthe form of increased MUC5AC-to-MUC5B ratios (22) andincreased DNA content (10). Despite the need for physiolog-ically relevant, healthy reference mucus, it is difficult to gen-erate sufficient material (mucus) to study intact mucus or thecomponents of mucus function in health or dysfunction indisease.

Currently, human airway mucus is sourced from inducedsputum (IS) in healthy subjects (2, 10, 21), spontaneous spu-tum from subjects with disease, or from human bronchialepithelial (HBE) cell culture surfaces (7, 19). Both “normal”and disease-associated sputum have strengths and weaknessesas sample types. Sputum induction in healthy subjects isrelatively low-risk and minimally invasive but has two limita-tions. First, induction dilutes mucus and, since rheology ishighly dependent on concentration (19), may perturb measuredviscoelastic properties. Second, induction requires a trainedhealth care team to perform the procedure and prepare thesample. In diseased patients, e.g., CF, sputum can be obtainedin large volumes (~5 mL) and often without induction (i.e.,“spontaneously”) (1, 5, 13). In certain diseased samples, spu-tum is subject to rapid proteolytic degradation of the mucinpolymer network and polymer-dependent mechanical proper-ties (21). Infection and inflammation also increase DNA con-

Address for reprint requests and other correspondence: D. B. Hill, MarsicoLung Inst., UNC, 125 Mason Farm Rd., Marsico Hall 7109, Chapel Hill, NC27599 (e-mail: david_b_hill@med.unc.edu).

centrations in diseased sputum (10), which, by its stiff nature(~10 times stiffer than mucins) (40, 42), may significantly altermucus rheological properties. CF sputum also has poor test-retest reliability in rheological assessments (37).

HBE cell culture is an effective way to obtain sterile,“healthy” airway mucus that is relatively free of proteolyticactivity, DNA, or other contamination. Furthermore, it is tun-able to various concentrations, allowing for the study of con-centration-dependent viscoelastic effects (19). However, HBEmucus does not contain the vascular compartment contribu-tions to mucus and is labor intensive to obtain, requiring thecare of a full-time laboratory technician to produce ~10 mL of2% solids mucus from 120 cultures of 1 million cells permonth.

In this study we investigated whether mucus obtained fromthe tips of endotracheal tubes (ETT) from normal subjects,without a history of lung disease or smoking, undergoingelective surgical procedures might be a practical source ofmucus for biophysical studies and biochemistry. ETT mucuscan be obtained in relatively large volumes (~0.6 mL per ETtube) as a by-product of outpatient surgery and, consequently,is readily available, and is not additionally invasive as acollection technique. While reports have appeared on ETTmucus (44), concerns about mucus dehydration due to evapo-ration (hypertonic dehydration) during intubation and surgerywere not fully addressed. Herein, we describe the feasibility ofcollecting native airway mucus from ETT and compare itsrheological and biochemical properties to those of HBE mucus,induced healthy sputum, and spontaneous CF sputum.

METHODS

Endotracheal tube collection and mucus recovery. ETTs werecollected by members of the Department of Anesthesiology at theUniversity of North Carolina (UNC) Hospital in Chapel Hill asapproved by the UNC Institutional Review Board (protocol no.11-0413). The tips (approximately the last 10 cm) of the ETTs wereremoved upon completion of surgical procedures. The tips wereplaced in a 50-mL conical tube, set on ice for transport to the UNCMarsico Lung Institute, and centrifuged at 400 g upon delivery toharvest the mucus. Each tip yielded ~600 �L of mucus (see Table 1and Supplemental Table S1; all Supplemental material is available at

https://doi.org/10.5281/zenodo.3351590). Samples were used imme-diately or snap-frozen and stored at �80°C to reduce proteolysis.

Collection of HBE mucus, induced healthy sputum, and spontane-ous CF sputum. HBE mucus was harvested from cell cultures andpooled as previously described (17). Briefly, HBE cells were grownon 1-cm2 Transwell inserts until confluent and mucus producing.Mature cultures were washed with phosphate-buffered saline (PBS),and washings were aspirated to remove surface debris and mucus.Mucus was dialyzed to a high-concentration working stock and frozenfor later use. Healthy induced (38) and CF (18) sputum samples werecollected under the approval of the University of North CarolinaInstitutional Review Board protocol no. 15-2431 and prepared forcharacterization and analysis as outlined previously (18). Demo-graphic information for each sample type is available in the respectivereferences above.

Measurement of mucus Na� and K� concentrations. Evaporativewater loss from ETT mucus can occur during intubation, even withventilation with nearly 100% humidified air, and during transport tothe laboratory. Because salt concentrations in normal mucus areisotonic with plasma, salt concentration can be used as an index ofmucus evaporative water loss (25). Additionally, mucus in healthyinduced sputum (IS) samples can be relatively dilute compared withthe airways due to salivary contamination. Because saliva has a verylow salt content (9), measuring Na� and K� concentrations can beused to estimate relative dilution of airway mucus during induction.Accordingly, 50-�L aliquots of each ETT and IS sample were usedfor determination of Na� and K� concentrations. Mucus sampleswere diluted into an equal volume of 6 N HCl-10% trichloroaceticacid and maintained at 70°C for 24 h (Optima, Fisher, Pittsburgh, PA).Following centrifugation, Na� and K� concentrations were measuredin the supernatant by inductive-coupled plasma optical emissionspectrometry (Optima 4300 DV, PerkinElmer, Norwalk, CT). Qualityassurance checks were obtained using a second multielement standard(SPEX CertiPrep, Metuchen, NJ). A multielement standard (VHGLaboratories, Manchester, NH) was used for the calibration curve.

Pooling and dilution of mucus for rheological characterizationacross concentrations. A “stock” ETT mucus (Table 2) was generatedby pooling 150 �L aliquots from the 15 isotonic ETT samplesdescribed in Table 1. The 15 samples were thawed, combined (vol-ume � 2.25 mL), and mixed by trituration in the presence of proteaseinhibitors (0.5 tablet; Roche Diagnostics, Mannheim, Germany) toprevent sample degradation. The stock was then mixed overnight ona rotator at 4°C. Aliquots of the pooled stock (4.6% solids) wereremoved, diluted, and mixed overnight at 4°C in PBS to generate 2%,3%, and 4% solids samples.

Table 1. Patient and sample characteristics for isotonic ETT mucus samples

Age, yr Sex Race Intubation Time, h Anesthesia Sample Volume, �L % Solids Na�, mM K�, mM Na/K Ratio 2(Na�K)

40 M Black 5.0 Isoflurane 700 4.84 121.08 30.98 3.91 30435 F Indian 1.5 Sevoflurane 550 10.04 110.52 23.40 4.72 26858 N Cauc 3.0 Isoflurane 550 3.45 123.39 47.19 2.61 34172 M Cauc 1.0 Sevoflurane 600 3.36 84.39 37.11 2.27 24358 F Cauc 2.0 Sevoflurane 600 3.39 117.67 29.37 4.01 29470 F Cauc 3.0 Isoflurane 650 5.45 120.53 27.07 4.45 29567 M Indian 6.0 Desflurane 450 9.03 143.53 21.09 6.81 32984 M Cauc 1.5 Isoflurane 1,150 3.66 135.55 26.60 5.10 32468 M Cauc 1.0 Sevoflurane 600 3.18 125.54 21.07 5.96 29321 M Black 1.5 Sevoflurane 550 6.26 121.86 45.90 2.65 33640 F Cauc 2.5 Sevoflurane 550 4.42 118.82 27.59 4.31 29347 M Cauc 2.0 Isoflurane 600 3.00 124.50 30.10 4.14 30937 F Cauc 2.0 Isoflurane 500 7.63 100.16 29.41 3.41 25943 M Black 2.0 Sevoflurane 750 3.15 132.49 32.95 4.02 33187 F Cauc 3.0 Sevoflurane 600 3.98 119.96 25.18 4.76 317

Mean 2.5 627 4.99 120.00 30.33 4.21 290

Cauc, Caucasian.

Determination of mucin concentrations in ETT and HBE mucusand sputum. A portion of the stock was reserved for biochemicalcharacterization of mucus concentration (% solids), mucin and proteinconcentrations, mucin molecular weight (MW), and mucin radius ofgyration (Rg) (16). In brief, ETT mucus samples were first solubilized1:1 in 6 M guanidinium HCl (hereinafter GuHCl; Fisher Scientific,Fairlawn, NJ) and, subsequently, diluted 1:200 in 200 mM sodiumchloride with 10 mM EDTA and 0.01% sodium azide (hereinaftercalled light-scattering buffer, i.e., LSB) to suit instrument sensitivities.The concentration and conformation of mucins, e.g., MW and Rg,were determined by gel permeation chromatography in series withmultiangle laser light scattering (MALLS) and differential refractom-etry. A 500 �L sample was added to a Sepharose CL-2B column (GEHealth Care Life Sciences) and eluted with LSB at a flow rate of 0.5mL/min. Effluents were passed through an in-line multiangle laserphotometer (DAWN Heleos, Wyatt Technology) coupled to a refrac-tometer (Trex, Wyatt Technology) to measure light scattering andsample concentration, respectively. The mass of mucins injected intothe system was determined by differential refractometry, whilemucin MW and Rg were determined by best fit of scattering data atmultiple angles to a second-order Berry model. Those values aregiven in Table 2.

HBE cells were prepared and cultured at an air-liquid interface, aspreviously described (26). The mucin and protein compositions of %solids concentration-matched mucus harvested from HBE cultureswere also measured. In parallel, a second, concentration-matched ETTpool was prepared from the same 15 ETT samples for comparisonwith HBE by MALLS, as described above. Total protein concentra-tions in ETT and HBE samples were determined by MALLS using adesalting column (GE Healthcare Life Sciences) with otherwise iden-tical methods, as described above (33).

Individual protein constituents of the ETT and HBE mucus weredetermined by mass spectrometry as outlined in the SupplementaryMethods (available at https://doi.org/10.5281/zenodo.3351590). Briefly,100 �L of the pooled ETT sample were denatured in 6 M GuHCl anddigested for assessment by mass spectrometry in a Dionex ultimate3000 RSLCnano system coupled to a hybrid quadrupole Orbitrapmass spectrometer with a nanospray source (Q-Exactive, ThermoFisher, Bremen, Germany). Proteins were identified by searchingagainst the most current human database (Proteome Discover 1.4) andquantified using Scaffold version 4 (Proteome Software) and thenormalized total precursor intensity.

Analysis of O-glycosylation of ETT mucin, sputum, and HBE cells.At 43 days in culture, HBE cells were washed with isotonic mediumand harvested by scraping. After pelleting, the cells were stored frozenat �80°C until analyzed. Frozen cell pellets were thawed, homoge-nized in ice-cold water, and extracted with organic solvents to removelipids and generate a lipid-free protein preparation for subsequentglycan analysis, as previously described (3, 4). Briefly, O-linkedglycans were released from glycoprotein preparations by reductive�-elimination, yielding intact, reduced structures. Following cleanup,the released O-linked glycans were permethylated and analyzed bynanospray ionization multidimensional mass spectrometry (NSI-MSn)

using instruments with linear and orbital ion trap capabilities (LTQ-Orbi, Thermo Fisher). Three biological replicates of cells culturedfrom the same donor were analyzed. Graphical representations ofmonosaccharide residues are presented in accordance with the broadlyaccepted Symbolic Nomenclature for Glycans, and O-linked glycananalysis was performed in keeping with the MIRAGE guidelines forglycomic studies (46, 49).

Rheology via cone-and-plate rheometer. Characterization of thebulk viscoelasticity of ETT mucus at designated concentrations wasperformed via cone (20 mm diameter, 1° deflection) and plate rhe-ometry. Forty microliters of mucus was loaded onto the Peltier plateof a DHR-3 rheometer (TA Instruments, New Castle, DE). Oscillatoryshear testing was performed over a range of small strain magnitudes(0.02–10% strain) to determine the linear viscoelastic regimen (LVR)for the elastic (G=) and viscous (G�) moduli. In the LVR, G= and G�are independent of shear strain, allowing for the frequency depen-dence of the viscoelastic moduli to be assessed across a range of0.1–10 radians/s (18). This methodology was used for all mucus typesstudied, with at least three technical replicates per sample type.

Particle-tracking microrheology. While bulk rheological measure-ments of G= and G� characterize the macroscopic properties of themucus gel, particle-tracking microrheology (PTMR) assesses themechanical properties of mucus at the length scale of its constituentbiopolymers (27, 44). In brief, the thermally driven motion of 1-�m-diameter carboxylated FluoSpheres (Thermo Fisher, Fremont, CA),henceforth referred to as “beads,” was tracked to determine theviscoelastic properties of the gel. Bead motion was recorded for 30 sat a rate of 60 frames/s using a 40 air objective on a Nikon EclipseTE2000U microscope. Individual bead trajectories were measuredautomatically using a custom Python program that uses TrackPy(https://doi.org/10.5281/zenodo.34028) for bead localization and track-ing. Bead motion was converted to mean squared displacement (MSD)and complex viscosity (*) values in accordance with the mathemat-ics described previously (19, 31).

Statistical methods. All statistical analyses were performed inMatlab (2017, MathWorks, Natick, MA). Differences in measuredvalues or trends were deemed statistically significant where P �� � 0.05. Differences between sample means were assessed by t-testusing the ttest2 function in Matlab, and differences in trends wereassessed by two-way ANOVA using the Matlab function anova2.

RESULTS

Selection and characterization of ETT samples for Na�, K�,and solids concentrations. A total of 77 samples were pro-cessed for % solids (see below), and dominant cation (Na� �K�) concentration measurements were made (SupplementalTable S1). The samples on average were evaporatively con-centrated (i.e., exhibited raised total Na� and K� concentra-tions relative to plasma), on average, by ~1.5-fold (see Sup-plemental Table S1). Isotonic ETT samples, i.e., samples withminimal evaporative water loss, were identified by selectingsamples with a Na� � K� value within 10% of the plasmavalue. All 15 isotonic samples identified were from nonsmok-ers without lung disease. Samples that were not within thenormal range of tonicity could theoretically be restored toisotonicity. However, the effects of evaporative hyperconcen-tration on mucus properties/organization and their reversibilitywith dilution are unknown, so attempts to dilute samples werenot undertaken.

The composition of minimal evaporative loss, i.e., isotonic,ETT mucus was compared with % solids and Na� and K�

concentrations of all collected ETT samples. Consistent withselection criteria, differences were present in Na� (Fig. 1A)and K� (Fig. 1B) concentrations between isotonic and whole-

Table 2. Characteristics of pooled endotracheal tube mucusstock used for rheology

Characteristic Pooled Value

Sample volume, �L 2,250% Solids 4.6[Na� � K�]mucus/[Na� � K�]plasma 1.01Mucins, �g/mL 3,720Molecular mass, kDa 2.42 109

Rg, nm 443.0

Plasma Na� and K� concentrations ([Na� � K�]) were taken to be 140 and4 mM, respectively. Rg, radius of gyration.

set samples (P � 0.001). Average sample volumes (Fig. 1C)and % solids values (Fig. 1D) were not significantly differentbetween the isotonic samples and the entire set.

Table 1 provides a detailed summary of the patient demo-graphics and composition of the isotonic samples. Mucus fromthese 15 subjects was collected with an average individual ETTsample volume of 647 �L. The mean ETT solids concentration(4.99%) is in excellent agreement with values reported from

tracheostomies by Matthews et al. (34) (i.e., 5.2%) and is inreasonable agreement with values for ETT mucus studied bySchuster et al. (44) (6.9%). The average Na� concentrationwas 120.00 13.60 mM and K� concentration was 30.86 7.86 mM.

Determination of pooled stock composition. The 2.25 mLETT stock pool had a % solids concentration of 4.6%, [Na� �K�]/[plasma Na� � K�] ratio of ~1, and total mucins con-centration of 3.72 mg/ml (see Table 2). The average molecularmass of the ETT mucin complexes measured by gel filtrationchromatography and MALLS was 2.42 109 Da with a radiusof gyration (Rg) of 443.0 nm. These values are similar to thosefor mucin complexes in CF sputum, with an average % solidsof 5.27 1.44%, a molecular mass of 2.4 0.57 109 Da,and Rg of 362 28 nm (n � 10). However, when mucins wereisolated by density gradient separation (36) in the presence ofa strong chaotropic agent, i.e., 6 M GuHCl, the molecular massof ETT mucus fell to 0.34 109 Da and 0.28 109 Da for CFsputum. Rg in these samples also decreased to 294 nm and 234nm for ETT and CF sputum, respectively. These values werecompared with the molecular mass of mucins isolated bydensity gradient from HBE cell culture, where molecular masswas 0.11 109 Da and Rg was 176 nm. Together, these dataindicate that mucus derived from both ETT and CF sputum invivo contains mucins assembled into complexes larger thanthose formed by cell cultures.

Proteomic analyses using mass spectrometry were employedto identify the other biomacromolecular constituents of ETTmucus. The top-20 protein species by intensity from the ETTmucus pool are listed in Table 3. Both MUC5B and MUC5ACwere detected, in addition to other airway-secreted proteins,such as lactotransferrin, deleted in malignant brain tumor 1protein, immunoglobulins, club cell secretory protein, proteinS100, and annexin from airway epithelium (14). Additionallyserum albumin and serotransferrin were present, indicating thatsome serum leak into these samples occurred.

Immunohistochemical staining of ETT mucus confirmed thepresence of mucins MUC5AC and MUC5B and identified

Fig. 1. Characteristics of all prospective endotracheal tube (ETT) mucussamples (All) and of those selected for being nearly isotonic (Iso). A and B:there were significantly lower (**P � 0.001) Na� (A) and K� (B) concentra-tions in the isotonic subset (Na�: 120.0 mM, K�: 30.3 mM) relative to theaverage of all samples (Na�: 197.9 mM, K�: 52.9 mM), indicating that somedehydration usually occurs during intubation. We also note that K� concen-tration in the isotonic samples was slightly increased above the 25 mMphysiological value (†P � 0.05). C: there was no discernable difference insample volume between the whole set of samples (581 �L) and the isotonicsubset (627 �L). D: there was also no difference in % solids between the wholeset and isotonic subset.

Table 3. Top-20 proteins by intensity as detected by mass spectrometry in pooled endotracheal tube mucus

Species Accession No. Alternate ID Molecular Mass, kDa Intensity

Serum albumin ALBU_HUMAN ALB 69 2.02E�12Lactotransferrin TRFL_HUMAN LTF 78 8.64E�11Lysozyme C LYSC_HUMAN LYZ 17 5.83E�11Immunoglobulin heavy constant-�1 IGHA1_HUMAN IGHA1 38 5.36E�11Immunoglobulin heavy constant-�2 A0A0G2JMB2_HUMAN IGHA2 37 4.79E�11Mucin-5B MUC5B_HUMAN MUC5B 596 4.38E�11Immunoglobulin heavy constant-�2 IGHA2_HUMAN IGHA2 37 3.44E�11Immunoglobulin-� constant IGKC_HUMAN IGKC 12 2.80E�11CCSP (uteroglobin) UTER_HUMAN SCGB1A1 10 2.34E�11Protein S100-A9 S10A9_HUMAN S100A9 13 2.25E�11Deleted in malignant brain tumors 1 protein DMBT1_HUMAN DMBT1 261 1.64E�11Polymeric immunoglobulin receptor PIGR_HUMAN PIGR 83 1.56E�11Mucin-5AC MUC5A_HUMAN MUC5AC 586 1.39E�11Actin, cytoplasmic 1 ACTB_HUMAN (�1) ACTB 42 1.29E�11BPI fold-containing family B member 1 BPIB1_HUMAN BPIFB1 52 1.21E�11Serotransferrin TRFE_HUMAN TF 77 1.15E�11Immunoglobulin �-constant 3 IGLC3_HUMAN IGLC3 11 8.79E�10Immunoglobulin heavy constant-�1 A0A0A0MS08_HUMAN (�1) IGHG1 44 8.67E�10Immunoglobulin �-like polypeptide 5 A0A0B4J231_HUMAN (�2) IGLL5 23 8.31E�10Annexin A1 ANXA1_HUMAN ANXA1 39 8.30E�10

CCSP, club cell secretory protein; BPI, bactericidal/permeability-increasing protein.

associated cell types (Supplemental Fig. S1). A combination ofdifferential interference contrast imaging and DNA stainingrevealed the presence of a number of cell types, includingpolymorphonuclear neutrophils, ciliated epithelial cells, andmacrophages (Supplemental Fig. S1, A and B). MUC5B ap-peared to be the dominant mucin compared with MUC5AC(Supplemental Fig. S1, C–F), which is in line with the massspectrometry results outlined in Table 3 and previous reports(22). Staining for fibrin (Supplemental Fig. S2B) was positive,despite limited detection of fibrin or fibrinogen by mass spec-trometry. A glandular source for at least some of the mucus/mucin in ETT samples was confirmed by staining for proline-rich 4 (PRR4) protein (Supplemental Fig. S2C).

Bulk rheology of pooled ETT mucus. We sought to deter-mine whether pooled ETT mucus exhibited rheologic proper-ties similar to those reported previously for respiratory mucus(44). We measured the LVR of the pooled stock over a span of0.2–10% strain for six separate 40-�L volumes of the pooledsample (Fig. 2, top). From those data, we selected 1% strain toassess the frequency-dependent viscoelastic moduli. Histori-cally, a frequency of 1 Hz has been used to compare values ofviscoelastic moduli, because 1 Hz is representative of muco-ciliary clearance (MCC) mechanics (21, 44). We determinedthat G= was always in excess of G� over the range of frequen-cies tested (Fig. 2, bottom). Further, we determined the G�-to-G= ratio at 1 Hz, i.e., tan(�), and report a value of 0.29. Thisvalue agrees well with previously reported values and sugges-tions that mucus obtained from the trachea is a cross-linkedgel (44).

Microrheology of ETT mucus. Thermally driven bead mo-tion was recorded in two dimensions for 30 s, generatingrepresentative trajectories, as shown in Fig. 3A. Of the fourmucus concentrations prepared by PBS dilution of the pool,only the beads in the 2% solids samples explored a spaceexceeding their diameter over the course of video recording.The time series of their displacements were converted intoMSDs. The ensemble mean log10(MSD) for each concentrationis shown in Fig. 3B, which illustrates the concentration depen-dence of bead motion and, hence, concentration dependence ofmucus viscoelasticity. MSD can be related to the lag time ofrecorded motion (�) via the fractional Brownian relationMSD � D���, where 0 � � � 1 is the subdiffusive exponentin the case of constrained thermal motion, as seen in entangled,elastic, and cross-linked gel systems (19, 28, 29). Only inpurely viscid, dissipative fluids is � � 1, and then D� � 4D,where D is the diffusion coefficient of the bead in its medium.The exponent � can be calculated as the slope of the curvelog10(MSD) vs. log10(�) for each sample, as shown in Fig. 3B.The exponent values are � � 0.55, 0.45, 0.25, and 0.14 for 2%,3%, 4%, and 4.6% solid samples, respectively. These dataindicate that all samples exhibited restricted, subdiffusive mo-tion and that particle motion was increasingly restricted withincreasing concentration.

Comparison of bulk and microrheological measurements.ETT mucus is believed to be isotropic but not homogeneous(44). While beads may, therefore, reliably report the viscoelas-

Fig. 2. Viscoelastic moduli of pooled 4.6% endotracheal tube (ETT) mucus.Top: strain-dependent viscoelastic moduli at 1 radian/s over a range of0.22–10.0% strain, which was determined to be the linear viscoelastic region.Bottom: frequency-dependent viscoelastic moduli at 1% strain over a range of0.016–1.6 Hz (0.1–10 radians). G= is in excess of G� over all frequencies,characteristic of a cross-linked gel.

Fig. 3. Measurement of particle motionover time for particle-tracking microrheol-ogy in endotracheal tube (ETT) mucus. A:particle motion was progressively hin-dered as mucus concentration increased;only in 2% mucus were particles able toprobe regions beyond their diameter (scalebar, 1 �m). B: ensemble means of particlemotion represented as mean squared dis-placement (MSD) vs. lag time (�). In log-log space, MSD curves are linear, andslopes for each mucus concentration wereless than 1 and decreased with concentra-tion, indicating subdiffusive motion andincreasingly hindered motion from interac-tions with the polymer mesh, respectively.Numbers of beads tracked per concentra-tion were 650, 474, 1,097, and 189 for 2%,3%, 4%, and 4.6%, respectively.

ticity of their local environment, they may not sense theviscoelasticity of the sample at all spatial scales or locales (47).Conversely, bulk measurements may not capture the localvariations in mesh size and cross-linking density at the mucin-polymer length scale. To test the features of each measurementtechnique, we compared PTMR with bulk measurements. Fig-ure 4 illustrates this comparison using the complex viscosity* (Pa·s) at 1 Hz as a combined measure of both the viscosityand the elasticity of ETT mucus over a range of mucusconcentrations selected for analysis. Mean bulk rheologicalmeasurements of * are shown as colored circles (means SE; Fig. 4, top). Microrheological measurements are shown asprobability density estimates of * for the ensemble of allbeads in a given sample. These estimates were generated usingthe Matlab (2018, MathWorks) function ksdensity.

Mean values of mucus complex viscosity as obtained fromboth methodologies are in good general agreement. The muchwider variability of * observed via PTMR indicates that thebeads are able to probe the random microstructural variationsof the samples. That the two methodologies generally agreeindicates that the size of the PTMR bead, i.e., slightly largerthan the mesh size, accurately sampled the heterogeneousmicrostructure of the mucus gel that imparts the macroscopicviscoelasticity of the gel.

Biochemical comparison of ETT mucus, HBE mucus, andnormal induced sputum. HBE mucus is an in vitro model forstudying the biophysics and biochemistry of airway mucus (18,19). However, questions have been raised regarding its simi-larity to native airway mucus and, therefore, its suitability as amodel system. We prepared a stock of HBE mucus for com-parison with a second pool of ETT mucus from the same 15subjects and measured the protein composition of ETT andHBE mucus using MALLS. These measurements were also

compared with the reported average values from 15 inducedsputum (IS) samples collected from healthy subjects as part ofa previous study (38). In general, measurements of the secondETT pool agreed well with the first pool results presented inTable 2. Mucin concentration was 4.44 mg/ml in the ETTpool and and 6.12 mg/ml in the HBE mucus pool (Fig. 5A),are similar to those reported by Matsui and colleagues inHBE mucus (33) and Schuster et al. in ETT mucus (44).Average mucin concentration was 1.64 mg/ml in the ISsamples (Fig. 5A).

Total nonmucin protein concentration was 11.8 mg/ml inETT mucus and 7.16 mg/ml in HBE mucus as calculated fromrefractometry using a desalting column. A desalting columnwas not used on IS samples as part of their analysis during theprevious study (38). Therefore, the value of total nonmucinprotein in IS samples was estimated as average solids content[1.2% solids, as previously reported (38)] minus mucin content(0.164%) and salt content, which was estimated from induc-tively coupled plasma atomic emission spectroscopy (ICP-OES) measurements of Na� and K� concentrations. Briefly,salt content was calculated from the following relationship:

�salt�IS ��Na��IS � �K��IS

�Na��ASL � �K��ASL

� �salt�ASL

�65.2 mM � 15.6 mM

120 mM � 25 mM� 0.9% � 0.5%

Accordingly, nonmucin protein concentration was deter-mined to be 0.53%, or 5.3 mg/ml, for IS. The total nonmucinprotein concentrations for each sample type are illustrated inlight gray in Fig. 5A. Figure 5B shows that ETT mucus had a

Fig. 4. Comparison of endotracheal tube (ETT) mucus bulk (macro) rheologyto particle-tracking microrheology (PTMR). Bulk rheological measurements(dots, means SE) of complex viscosity (*) agree well with the centraltendencies of the PTMR ensemble distributions (filled regions) of * in allmucus concentrations. PTMR * distributions are represented as kernel-smoothed estimates of the probability density. The ordinate value chosen forbulk measurements was arbitrary.

Fig. 5. Mucin and protein composition in endotracheal tube (ETT), inducedsputum (IS), and human bronchial endothelial (HBE) mucus via multianglelaser light scattering (MALLS). A: total solids composition of ETT, IS, andHBE mucus as % salts (0.9% ETT vs. 0.9% HBE vs. 0.5% IS), mucins (0.44%ETT vs. 0.61% HBE vs. 0.164% IS), and other proteins (1.18% ETT vs. 0.72%HBE vs. 0.53% IS). B: mucin-to-other protein ratio in ETT (0.27), IS (0.23),and HBE (0.46). C: molecular mass of mucins in ETT (2.28 109 Da), IS(5.31 108 Da), and HBE (4.10 108 Da). D: radius of gyration (Rg) ofmucins in ETT (440.9 nm), IS (351.7 nm), and HBE (171.9 nm) mucus. Errorbars show means SD.

mucin-to-total protein ratio of 0.27 that is roughly twofoldlower than the 0.46 of HBE mucus. In contrast, IS had amucin-to-total protein ratio of 0.23, or roughly equal to that ofETT.

The molecular mass and size of ETT samples analyzed byMALLS without prior density gradient centrifugation werealso compared with values for IS and HBE. The molecularmass of the mucus peak components of ETT samples (2.28 109 0.2 109 Da, means SD of both pools) was roughlyfour to five times greater than that of IS (5.3 4.0 108 Da,n � 15 historical samples) or HBE (4.1 1.6 108 Da, n �8 historical samples) (Fig. 5C). This difference was associa-ted with an increase in radius of gyration for ETT mucus(440.9 2.9 nm) vs. IS (351.7 60.4 nm) or HBE (171.9 25.2 nm) mucus (Fig. 5D).

Taken together, these findings indicate that the concentra-tions of mucin in ETT and HBE mucus are roughly similar.However, the total protein content is higher in ETT mucus,and the mucin percentage of total protein lower, reflectingaddition of serum and, likely, general airway epithelialproteins in ETT samples. Sputum induction is known todilute samples (16) and is associated with salivary contam-ination, which may explain the relatively low mucin con-centrations in the IS samples. That the relative proportion ofmucins among total proteins is preserved by correcting forsalivary dilution based on salt concentration indicates thatsalivary contamination likely occurred.

Glycosylation of ETT, HBE, and CF sputum samples. Gly-cosylation of the mucin VNTR regions can be a key determi-nant of mucin properties. We compared the O-linked glycanprofiles of ETT mucus (n � 3) with the profiles of HBE (n �3) and IS (n � 3). Among the most abundant O-linked glycans,13 nonsulfated and 8 sulfated structures were shared across thethree sample types. These structures, as well as less abundantglycans that were detectable but not quantifiable, demonstratedthe expected heterogeneity arising from normal human geneticvariation in the fucosyltransferase encoded at the secretor (sec)locus. This heterogeneity was detected as increased addition offucose (Fuc) to glycans harvested from sec�/� individuals andincreased addition of sialic acid (NeuAc) in sec�/� individuals.Glycans harvested from sec�/� individuals exhibit intermedi-ate relative abundances of NeuAc and Fuc addition (35).

Samples of ETT and IS mucus were obtained from all threesecretor genotypes, and HBE cells were prepared in threebiological replicates from sec�/� subjects. The relative abun-dance of nonsulfated glycans was compared across samples byhierarchical clustering (49) and demonstrated that sec�/� sta-tus segregated the glycan profiles of ETT, sputum, and HBEaway from the other two genotypes (Fig. 6, left). Thus, non-sulfated glycan profiles distinguished the secretor status of thedonor more than the source of the material (ETT, IS, HBE).However, for sulfated glycans, ETT and HBE profiles segre-gated away from healthy IS samples regardless of secretorstatus, and hierarchical clustering of sulfated glycans failed todetect differences between HBE cells and ETT mucus (Fig. 6,right); sulfated glycan profiles indicated greater similaritybetween ETT mucus and HBE cells than between HBE/ETTmucus and IS. Furthermore, this difference was observedindependent of the secretor status of the donor (24, 30, 43). Themajor factor driving the clustering of IS away from HBE/ETTwas the reduced structural complexity of the sulfated glycans

in IS compared with HBE/ETT. The relative abundance of asingle glycan (sulfated glycan #5) dominates the IS profilecompared with the greater diversity of sulfated structuresdetected in HBE and ETT (Fig. 6, right). In HBE and ETT,sulfated glycan #5 is also highly abundant but shares the statusof highest relative abundance with sulfated glycan #3, thesialylated form of sulfated glycan #5. Reduced relative abun-dance of sialylated glycans in IS was also detected in thenonsulfated glycans (Fig. 6, left). The three most abundantsialylated, nonsulfated glycans (nonsulfated structures #5, 6,and 7, Fig. 6, left) accounted for 34 13, 48 8, and 77 2%(means SE) of the total nonsulfated glycan profile for IS,ETT, and HBE, respectively. Thus, in comparison to inducedsputum, ETT and HBE mucus is characterized by a higherrelative abundance of sialylated and sulfosialylated glycans.

Biophysical comparison of ETT mucus with HBE mucus.Because increased mucus concentration (i.e., isotonic dehydra-tion) is a primary pathophysiologic variable in MOLDs, suchas CF (2, 7, 18, 19, 33), we sought to determine whether theconcentration dependence of mucus viscoelasticity was similarin ETT and HBE mucus. Hill and colleagues (19) previouslyperformed a concentration-dependent microrheological analy-sis of HBE mucus over a range of concentrations similar tothose utilized in the present study. Figure 7A demonstratesclose agreement for * in ETT (blue curves) and HBE (orangecurves) mucus across this range of concentrations. Two-wayANOVA revealed that concentration had a significant effect onchanging *, while sample type did not. Additionally, a nearlyidentical scaling behavior as seen in ETT and HBE mucus wasfound in sputum samples from individual patients with CF(Supplemental Fig. S3).

Frequency responses as measured by PTMR in ETT vs.HBE mucus illustrated that while agreement was not perfectbetween sample types at 2% (Fig. 7B), 3% (Fig. 7C), and 4%(Fig. 7D), it was better than within concentrations for eachsample type, as determined at 1 Hz by two-way ANOVA (P �0.05 for % solids vs. P � not significant for sample type). Putsimply, concentration had more influence than sample originon viscoelastic properties. This conclusion was also borne outat 0.1 and 10 Hz (see Supplemental Fig. S4). These findingsindicate that both sample types may be useful for studying theviscoelastic mechanics of healthy airway mucus across a rangeof concentrations. Notably, the loss modulus G� was greaterthan G= in all samples except 4% solids ETT, in line with dataHill and colleagues had previously reported in HBE regardinga sol-gel transition occurring near that concentration (19).

DISCUSSION

To study the rheologic properties of the mucus that normallylines airway surfaces, an abundant, minimally invasive, andrepresentative source of airway mucus is needed. Endotrachealtube mucus has been analyzed as a source of airway mucus ina few studies (10, 44) and compared with other sample typesand model systems. In this work, we expand on previousreports by performing a detailed biochemical and biophysicalanalysis of ETT mucus obtained from healthy subjects underconditions designed to mimic the native airway environment.

Our studies established that, from a total of 77 samples, wecould identify 15 separate isotonic patient ETTs that, whencombined, produced a stock with a volume greater than the

volume generated from 1 mo of harvesting HBE mucus from�100 HBE cultures (on 1 cm2 inserts). Human bronchialepithelial cells have been a gold standard for mucus produc-tion, because HBE cultures recapitulate many of the compo-nents of the MCC mechanism in airways, including secretorycell mucus production and coordinated, mucomotive ciliarybeat. For this reason, despite the labor-intensive requirement toproduce small mucus volumes, HBE cultures have been widelyused to study MOLDs (17, 19, 20, 33). However, questionsremain as to whether the absence of gland mucus and thenormal plasma proteins that passively permeate into mucuslimit HBE mucus as being representative of in vivo airwaymucus. Induced sputum from healthy subjects is minimallyinvasive, but inhalation of hypertonic saline (HS) for inductionis time/medical personnel-consuming. Notably, HS inductionmay dilute mucus by ~25% (15), and salivary contaminationgenerates a second form of dilution.

In addition to the relative ease of obtaining ETT mucus, ETTmucus is attractive as a source of mucus because it is an “invivo”-produced airway mucus. ETT mucus can also be ob-tained in relatively large (�0.6 mL/ETT; Fig. 1A) volumes

multiple times per week as a by-product of surgeries performedon subjects without respiratory disease. These sample volumeswould be sufficient to use with methods that require largeraliquots [e.g., �3 �L for flame photometry (25)] to accuratelyassess ion concentration for those without access to ICP-OESequipment. Typical ETT samples, on average, were composedof ~5.6 2.8% solids (Fig. 1, bottom right) and exhibited Na�

and K� concentrations that were ~1.5-fold greater than ASLand plasma values (Fig. 1, bottom left). This hypertonicity islikely explained by evaporative water loss during mechanicalventilation. However, 15 of 77 samples exhibited sums ofcations, i.e., Na� and K�, that were within 10% of plasma Na�

and K� and, hence, were isotonic with minimal evaporativewater loss (Table 1). We pooled these samples for studies ofETT mucus to avoid potential artifacts associated with samplesthat experienced evaporation and would require water dilutionto restore isotonicity (Table 2). Furthermore, salt concentrationhas been shown to have a significant effect on the biophysicalbehavior of mucin gels (47), underscoring the utility of asample type with physiological electrolyte composition.

Fig. 6. Hierarchical clustering of O-linked glycan profiles of endotracheal tube (ETT) mucin, sputum (S), and human bronchial epithelial (HBE) cells. Left:nonsulfated O-linked glycans were harvested and analyzed by nanospray ionization multidimensional mass spectrometry (NSI-MSn) in positive-ion mode. The13 highest-abundance O-linked glycans detected in ETT mucus (ETT), sputum (S) from healthy subjects, and human bronchial epithelial cells cultured at anair-liquid interface (HBE) were quantified as the percentage of the signal intensity that each individual glycan contributed to the total signal intensity for all 13glycans (percentage of total profile). Based on the total glycan profile, donors of the indicated samples were characterized as sec�/�, sec�/�, or sec�/� for thefucosyltransferase activity encoded at the human secretor (sec) locus. Hierarchical clustering was performed to assess the closest similarities across the samples.Sec�/� samples segregated together, regardless of their source (ETT, S, or HBE), indicating that secretor status outweighs sample source as a determinant ofnonsulfated glycosylation. Right: sulfated O-linked glycans were harvested and analyzed by NSI-MSn in negative-ion mode. The relative abundance of the 7 mostabundant sulfated O-linked glycans detected in all three sample types were quantified and compared with each other as percentage of total profile. Hierarchicalclustering of sulfated glycan abundance demonstrates greater similarity between ETT mucus and HBE cells than between ETT mucus and sputum. Thesimilarities in sulfated glycan profiles are independent of secretor status. Graphical representations of monosaccharide residues are in accordance with the broadlyadopted Symbolic Nomenclature for Glycans guidelines (SNFG).

Cone-and-plate rheometry revealed that mucus from thepooled ETT stock exhibited viscoelastic properties that werevirtually identical to the average of previously reported indi-vidually tested ETT samples (41, 44). Figure 2 illustrates thatthe pooled sample behaved as a viscoelastic, cross-linked gel,and computation of tan(�) at 1 Hz (0.29) revealed viscous andelastic properties close to tan(�) � 0.3 reported by Schuster etal. (44) and tan(�) � 0.33 reported by Rubin et al. (41) for ETTmucus. This data set suggests that pooling ETT mucus intolarge volumes did not alter the bulk rheological behavior ofnative mucus and offers the advantage of providing for largerworking volumes of airway mucus for well-controlled mac-rorheological analyses.

PTMR is useful in the analysis of entangled and cross-linkedgels, because it can reveal the viscous and elastic properties ofthe gel at the scale of its polymer network (48). Hindered (i.e.,non-Brownian, subdiffusive) bead motion is related to theelasticity of the gel, while diffusion is reflective of the dissi-pative properties of the interstitial liquid within the mesh.These two elements can be visualized in Fig. 3A, where, at 2%solids concentrations, the polystyrene beads were able to movebeyond their own diameter over the time course of the exper-iment. At 3% solids concentration, bead motion was “caged” toan area within its own diameter. At higher solids concentra-tions, bead motion was severely constrained. This caging is

also illustrated in Fig. 3B, where the base-10 logarithm of themean squared displacements of bead motion are plotted againstthe logarithm of lag time, �. In purely viscid liquids, the slopeof the log-transformed curve is � � 1. As elastic interactionsbetween beads and mucus increase, the slope decreases as beadmotion became more restricted. The concentration-dependentdecrease in MSD in Fig. 3B is reflective of the concentration-dependent increase in mucus gel stiffness that contributes tofailure of MCC in MOLDs.

PTMR is also an important methodology for the study of thebiophysical properties of mucus, because it is able to capturethe innate heterogeneity of the fluid on the length scale of itsmolecular constituents. By evaluating the rheology explored byan ensemble of beads in a sample type, the “rheologicaltopology” of a sample can be determined, sometimes revealingemergent behaviors compared with bulk measurements. Figure4 illustrates this feature, where the mean viscoelastic characterof the mucus was similar for bulk and PTMR measurements.However, the PTMR measurements were more heterogeneous.In particular, the probability density estimate of the 2% solidsensemble reveals rheologies across nearly 3 logs of * and aclear bimodality. However, the central tendency of the ensem-ble PTMR data was in agreement with bulk mean values,suggesting that bulk rheological measurements provide acoarse-grain survey of the biophysical properties of a sample.

The bimodality of the 2% sample is notable for the similarityof its higher-viscosity mode to the distribution of the 3%sample. As these samples were obtained via serial dilution, itmay be that the two modes represent a snapshot of a transitionbetween phases. Georgiades and colleagues (12) reported thatthe transition between entanglement and overlap in porcinegastric and duodenal mucus was between 2% and 3% solids,respectively. It may be that our samples were not fully mixedinto the PBS diluent to form a uniform phase.

Relevant biochemical characterizations of ETT mucus arethe absolute mucin concentration and the ratio of mucin to“other” biologic molecule composition and how those valuescompare with HBE mucus and IS from normal subjects. Figure5A illustrates the total solids content of ETT, IS, and HBEmucus. Mucin complexes comprised slightly more than 0.5%of both ETT and HBE samples, with ETT mucin complexesmeasured at 4.44 mg/ml and HBE mucins at 6.12 mg/ml, whileIS samples had lower mucin concentrations of 1.64 mg/ml.Differences were also observed in the quantity of nonmucinproteins in the samples. ETT mucus (11.8 mg/ml) had abso-lutely more nonmucin protein material than IS (5.35 mg/ml)and HBE mucus (7.16 mg/ml).

Figure 5B illustrates the differences in mucin-to-other pro-tein ratios between ETT, IS, and HBE samples. The proteomicsdata identified the source of increased nonmucin proteins inETT versus HBE as serum leak proteins (albumin) and airwayepithelial proteins that may be glandular in origin. Thus, ETTsmay sample a broader variety of proteins that normally are partof mucus in vivo than HBE cells. The relatively similarproportions of mucin and nonmucin proteins may be explained,at least in part, by the dilution of both mucins and nonmucinprotein by HS induction. The possible contribution of salivarycontamination to IS mucin/nonmucin protein concentrations isunknown. The data also indicate that rough comparisons ofmucin concentrations between in vivo samples and/or HBEsamples based on % solids may be inaccurate.

Fig. 7. Comparison of particle-tracking microrheology measurements of vis-coelasticity in endotracheal tube (ETT, blue) and human bronchial endothelial(HBE, orange) mucus. A: complex viscosity (*) in ETT and HBE mucus.Slopes of lines of best fit for HBE and ETT excluding 2% were m � 5.2 andm � 4.9, respectively, similar to values reported by Georgiades and colleagues(12) for porcine duodenal mucus. Best-fit slope for all ETT concentrations was3.7, similar to Georgiades et al. report of 3.9 in porcine gastric mucus. B:viscous modulus (G�) exceeded the loss modulus (G=) for both 2% HBE andETT mucus, characteristic of a sol. Both measures agreed well between sampletypes. C: general agreement was also observed for G= and G� between sampletypes at 3% solids with G� again exceeding G= in ETT and HBE samples. D:values G= and G� were near to each other in both 4% solids ETT and HBEmucus, indicating proximity to the sol-gel transition point, with similaritiesbetween sample types as well. However, G= exceeded G� in ETT indicatingthat ETT mucus had undergone transition to a gel, whereas Hill et al. (19)previously reported that HBE mucus does not undergo this transition until %solids is just above 4%. [All HBE data reproduced from Hill et al. (19), withpermission under the Creative Commons Attribution 4.0 Unported License.]

We also performed MALLS with pooled ETT mucus forcomparison with HBE and IS samples with nominally identical% solids content. Figure 5, C and D, illustrates that the starkestdifference between ETT mucus, IS, and HBE mucus was in thesize of their mucin complexes. ETT mucins formed enormouscomplexes with molecular mass of 2.28 GDa (Fig. 5C) and Rg

of 440.9 nm (Fig. 5D), respectively. In contrast, HBE mucinsformed smaller complexes with molecular mass of 410.0 MDaand Rg of 171.9 nm. IS was associated with smaller complexesbut similar Rg. When mucins from all sample types wereexposed to isopycnic centrifugation, molecular mass of ETTmucus fell and was similar to values reported for IS and HBEmucus, with molecular mass of 338 MDa. The large complexesfound in ETT mucus before chaotropic, isopycnic isolationmay be formed due to the presence of mucin interactions withglobular proteins, aggregation of mucins with inhaled particles,or large submucosal gland extrusions (11) not found in HBEcultures. Furthermore, because the IS samples appeared tosuffer from salivary contamination, as evidenced by dilute ionconcentrations relative to ASL values, the IS molecular massand Rg values may reflect a component of salivary mucincomplexes that have been reported to have molecular mass ofroughly 10–40 MDa and proportionate Rg values (39). To-gether, these data indicate that studies of large-molecular-weight macromolecular forms of mucins may be best per-formed in ETT samples.

The hydrophilic nature and anionic charge of the glycansattached to mucins contribute significantly to the physicochem-ical properties of mucus. Our glycomics analysis demonstratedthat ETT and HBE mucus were more similar to each other thanto IS from healthy individuals (Fig. 6). The O-linked glycans ofETT and HBE, in comparison to IS, exhibited greater relativeabundance of nonsulfated sialylated glycans and of sulfatedsialylated glycans. These glycomic signatures may reflect in-teresting, but currently uncharacterized, differential expressionof specific sulfo- and sialyltransferases in the secretory epithe-lia from which the mucus is derived. Alternatively, sputumglycans may be susceptible to desialylation by host salivary ormicrobial sialidases that do not have access to O-linked gly-cans harvested through ETT or are not present in the asepticenvironment of HBE cultures. Our observation that the majorO-linked glycans of ETT mucus were more similar to HBEmucin glycans than IS glycans suggests that IS may notappropriately reflect the characteristics of airway mucus and, tothe extent that glycosylation influences mucus function, maynot be a representative sample type for study of the physico-chemical properties of airway mucus.

Using PTMR, we were able to compare the rheology of ETTmucus with HBE mucus. Figure 7A illustrates that the com-posite viscoelasticity (as *) was similar for each sample typeand that each displayed a similar dependence on concentration.Viscosity has a power law dependence on concentration thatcan be calculated from the slope (m) of the curve log10(*) vs.log10(c). Our reported values for HBE (m � 5.2) and ETTmucus above 2% solids (m � 4.9) agree well with valuesreported by Georgiades et al. in porcine duodenal mucus underentanglement regimens (12). If concerns regarding the phase(i.e., overlap vs. entanglement) implications of the bimodalityin 2% ETT mucus are disregarded, adding its reported viscos-ity to the fit yields m � 3.74, which agrees with the valuesreported by Georgiades et al. for porcine gastric mucus, where

m � 3.92. Additionally, the behavior of these pooled samplesagreed well with data from individual CF sputum samples(Supplemental Fig. S3).

We also used PTMR to compare the frequency responses ofETT and HBE mucus. Values for G= and G� were in goodagreement for both sample types at 2% solids concentration,particularly at higher frequencies (Fig. 7B). Generally goodagreement was also found between ETT and HBE mucus at 3%solids (Fig. 7C). In all 2% and 3% measurements, G� exceededG=, which is characteristic of a viscoelastic sol (19). At 4%solids, G= and G� were similar in ETT mucus, as has beenreported previously by Hill and colleagues (19). In contrast toprevious and current (Fig. 7A) HBE data, G= exceeded G� inETT mucus of 4% solids concentration. This behavior ischaracteristic of a viscoelastic gel, implying that between 3%and 4% solids, ETT mucus undergoes a sol-gel transition,whereas HBE mucus undergoes a similar transition at justabove 4% solids (19). Despite this difference between the twosample types, ETT samples provide a source of native airwaymucus that has both sol and gel phases and viscoelasticbehavior similar to other mucus types.

In conclusion, this study describes extended biochemicaland biophysical analyses of native human airway mucus ob-tained from endotracheal tubes (ETTs). We selected and com-bined mucus obtained only from ETTs with salt concentrationssimilar to those found in the airway surface liquid, i.e., withminimal evaporative water loss. A large volume of pooled ETTmucus samples behaved rheologically similarly to previousreports of individual ETT samples. The pooled samples weretunable with respect to mucus concentration while maintainingphysiological (isotonic) Na� and K� concentrations. Our bio-physical measurements agreed well with previous studies inother mucus types, including mucus obtained from HBE cellcultures in this and previous work, where sample volumes canbe prohibitively small. ETT mucus may have a broader repre-sentation of endogenous nonmucin organic molecules that maycontribute to mucin organization in vivo. In summary, ETTmucus is a viable source of large volumes of native, tunableairway mucus for the study of muco-obstructive airway dis-eases.

ACKNOWLEDGMENTS

The authors thank Dr. Brian Button for help in measuring individual sample% solids content and Stephanie Livengood for assistance with MALLS.

GRANTS

This work was funded by Cystic Fibrosis Foundation Grants SUBRAM17I0,MARKOV18F0, EHRE16XX0, Hill16XX0, and Bouche15R0, National Instituteof Allergy and Infectious Diseases Grant AI-110098, and National Institute ofDiabetes and Digestive and Kidney Diseases Grant P30DK065988.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.R.M., D.B.S., K.A.R., H.A., P.A.K., D.B.N., T.K., T.M.K., D.C.K.,R.C.B., C.E., and D.B.H. conceived and designed research; M.R.M., W.J.K.,C.B.M., I.C.G., A.L.G., K.A.R., D.B.N., T.K., T.M.K., G.R., and N.E.A.performed experiments; M.R.M., D.B.S., W.J.K., C.B.M., I.C.G., A.L.G.,K.A.R., D.B.N., T.K., T.M.K., D.C.K., G.R., M.T., C.E., and D.B.H. analyzeddata; M.R.M., D.B.S., A.L.G., K.A.R., P.A.K., D.B.N., T.K., G.R., N.E.A.,M.K., M.T., R.C.B., C.E., and D.B.H. interpreted results of experiments;M.R.M., C.B.M., T.K., and M.T. prepared figures; M.R.M. drafted manuscript;

M.R.M., D.B.S., W.J.K., C.B.M., I.C.G., K.A.R., H.A., P.A.K., N.E.A., M.K.,M.T., R.C.B., C.E., and D.B.H. edited and revised manuscript; M.R.M.,D.B.S., W.J.K., C.B.M., I.C.G., A.L.G., K.A.R., H.A., P.A.K., D.B.N., T.K.,T.M.K., D.C.K., G.R., N.E.A., M.K., M.T., R.C.B., C.E., and D.B.H. approvedfinal version of manuscript.

REFERENCES

1. Alexis NE, Wells H, Siperly E, Goldstein B, Henderson AG, Peden DB.Baseline sputum parameters in normals, asthmatics, COPD, atopics, smok-ers and ex-smokers (Abstract). J Allergy Clin Immunol 137: AB208, 2016.doi:10.1016/j.jaci.2015.12.1121.

2. Anderson WH, Coakley RD, Button B, Henderson AG, Zeman KL,Alexis NE, Peden DB, Lazarowski ER, Davis CW, Bailey S, Fuller F,Almond M, Qaqish B, Bordonali E, Rubinstein M, Bennett WD,Kesimer M, Boucher RC. The relationship of mucus concentration(hydration) to mucus osmotic pressure and transport in chronic bronchitis.Am J Respir Crit Care Med 192: 182–190, 2015. doi:10.1164/rccm.201412-2230OC.

3. Aoki K, Perlman M, Lim JM, Cantu R, Wells L, Tiemeyer M.Dynamic developmental elaboration of N-linked glycan complexity in theDrosophila melanogaster embryo. J Biol Chem 282: 9127–9142, 2007.doi:10.1074/jbc.M606711200.

4. Aoki K, Porterfield M, Lee SS, Dong B, Nguyen K, McGlamry KH,Tiemeyer M. The diversity of O-linked glycans expressed during Dro-sophila melanogaster development reflects stage- and tissue-specific re-quirements for cell signaling. J Biol Chem 283: 30385–30400, 2008.doi:10.1074/jbc.M804925200.

5. Bhowmik A, Seemungal TA, Sapsford RJ, Devalia JL, Wedzicha JA.Comparison of spontaneous and induced sputum for investigation ofairway inflammation in chronic obstructive pulmonary disease. Thorax 53:953–956, 1998. doi:10.1136/thx.53.11.953.

6. Button B, Anderson WH, Boucher RC. Mucus hyperconcentration as aunifying aspect of the chronic bronchitic phenotype. Ann Am Thorac Soc13, Suppl 2: S156–S162, 2016. doi:10.1513/AnnalsATS.201507-455KV.

7. Button B, Cai LH, Ehre C, Kesimer M, Hill DB, Sheehan JK, BoucherRC, Rubinstein M. A periciliary brush promotes the lung health byseparating the mucus layer from airway epithelia. Science 337: 937–941,2012. doi:10.1126/science.1223012.

8. Casalino-Matsuda SM, Monzón ME, Forteza RM. Epidermal growthfactor receptor activation by epidermal growth factor mediates oxidant-induced goblet cell metaplasia in human airway epithelium. Am J RespirCell Mol Biol 34: 581–591, 2006. doi:10.1165/rcmb.2005-0386OC.

9. Dawes C. Circadian rhythms in human salivary flow rate and composition.J Physiol 220: 529–545, 1972. doi:10.1113/jphysiol.1972.sp009721.

10. Duncan GA, Jung J, Joseph A, Thaxton AL, West NE, Boyle MP,Hanes J, Suk JS. Microstructural alterations of sputum in cystic fibrosislung disease. JCI Insight 1: e88198, 2016. doi:10.1172/jci.insight.88198.

11. Ermund A, Meiss LN, Rodriguez-Pineiro AM, Bähr A, Nilsson HE,Trillo-Muyo S, Ridley C, Thornton DJ, Wine JJ, Hebert H, KlymiukN, Hansson GC. The normal trachea is cleaned by MUC5B mucinbundles from the submucosal glands coated with the MUC5AC mucin.Biochem Biophys Res Commun 492: 331–337, 2017. doi:10.1016/j.bbrc.2017.08.113.

12. Georgiades P, Pudney PDA, Thornton DJ, Waigh TA. Particle trackingmicrorheology of purified gastrointestinal mucins. Biopolymers 101: 366–377, 2014. doi:10.1002/bip.22372.

13. Gershman NH, Wong HH, Liu JT, Mahlmeister MJ, Fahy JV.Comparison of two methods of collecting induced sputum in asthmaticsubjects. Eur Respir J 9: 2448 –2453, 1996. doi:10.1183/09031936.96.09122448.

14. Fischer AJ, Goss KL, Scheetz TE, Wohlford-Lenane CL, Snyder JM,McCray PB Jr. Differential gene expression in human conducting airwaysurface epithelia and submucosal glands. Am J Respir Cell Mol Biol 40:189–199, 2009. doi:10.1165/rcmb.2008-0240OC.

15. Henderson AG, Anderson WH, Ceppe A, Coakley RD, Button B,Alexis NE, Peden DB, Lazarowski ER, Davis CW, Fuller F, AlmondM, Qaqish B, Kesimer M, Boucher RC. Mucus hydration in subjectswith stable chronic bronchitis: a comparison of spontaneous and inducedsputum. COPD 15: 572–580, 2018. doi:10.1080/15412555.2019.1566892.

16. Henderson AG, Ehre C, Button B, Abdullah LH, Cai LH, Leigh MW,DeMaria GC, Matsui H, Donaldson SH, Davis CW, Sheehan JK,Boucher RC, Kesimer M. Cystic fibrosis airway secretions exhibit mucinhyperconcentration and increased osmotic pressure. J Clin Invest 124:3047–3060, 2014. doi:10.1172/JCI73469.

17. Hill DB, Button B. Establishment of respiratory air-liquid interfacecultures and their use in studying mucin production, secretion, andfunction. In: Mucins: Methods and Protocols, edited by McGuckin MA,Thornton DJ. New York: Springer, 2012, p. 245–258.

18. Hill DB, Long RF, Kissner WJ, Atieh E, Garbarine IC, MarkovetzMR, Fontana NC, Christy M, Habibpour M, Tarran R, Forest MG,Boucher RC, Button B. Pathological mucus and impaired mucus clear-ance in cystic fibrosis patients result from increased concentration, notaltered pH. Eur Respir J 52: 1801297, 2018. doi:10.1183/13993003.01297-2018.

19. Hill DB, Vasquez PA, Mellnik J, McKinley SA, Vose A, Mu F,Henderson AG, Donaldson SH, Alexis NE, Boucher RC, Forest MG.A biophysical basis for mucus solids concentration as a candidate bio-marker for airways disease. PLoS One 9: e87681, 2014. [Erratum in PLoSOne 9(5): e97980, 2014.] doi:10.1371/journal.pone.0087681.

20. Holmén JM, Karlsson NG, Abdullah LH, Randell SH, Sheehan JK,Hansson GC, Davis CW. Mucins and their O-glycans from humanbronchial epithelial cell cultures. Am J Physiol Lung Cell Mol Physiol 287:L824–L834, 2004. doi:10.1152/ajplung.00108.2004.

21. Innes AL, Carrington SD, Thornton DJ, Kirkham S, Rousseau K,Dougherty RH, Raymond WW, Caughey GH, Muller SJ, Fahy JV. Exvivo sputum analysis reveals impairment of protease-dependent mucusdegradation by plasma proteins in acute asthma. Am J Respir Crit CareMed 180: 203–210, 2009. doi:10.1164/rccm.200807-1056OC.

22. Kesimer M, Ford AA, Ceppe A, Radicioni G, Cao R, Davis CW,Doerschuk CM, Alexis NE, Anderson WH, Henderson AG, Barr RG,Bleecker ER, Christenson SA, Cooper CB, Han MK, Hansel NN,Hastie AT, Hoffman EA, Kanner RE, Martinez F, Paine R III,Woodruff PG, O’Neal WK, Boucher RC. Airway mucin concentrationas a marker of chronic bronchitis. N Engl J Med 377: 911–922, 2017.doi:10.1056/NEJMoa1701632.

23. Kesimer M, Kirkham S, Pickles RJ, Henderson AG, Alexis NE,Demaria G, Knight D, Thornton DJ, Sheehan JK. Tracheobronchialair-liquid interface cell culture: a model for innate mucosal defense of theupper airways? Am J Physiol Lung Cell Mol Physiol 296: L92–L100,2009. doi:10.1152/ajplung.90388.2008.

24. Kiwamoto T, Katoh T, Evans CM, Janssen WJ, Brummet ME,Hudson SA, Zhu Z, Tiemeyer M, Bochner BS. Endogenous airwaymucins carry glycans that bind Siglec-F and induce eosinophil apoptosis.J Allergy Clin Immunol 135: 1329–1340.e9, 2015. doi:10.1016/j.jaci.2014.10.027.

25. Knowles MR, Robinson JM, Wood RE, Pue CA, Mentz WM, WagerGC, Gatzy JT, Boucher RC. Ion composition of airway surface liquid ofpatients with cystic fibrosis as compared with normal and disease-controlsubjects. J Clin Invest 100: 2588–2595, 1997. doi:10.1172/JCI119802.

26. Krunkosky TM, Jordan JL, Chambers E, Krause DC. Mycoplasmapneumoniae host-pathogen studies in an air-liquid culture of differentiatedhuman airway epithelial cells. Microb Pathog 42: 98–103, 2007. doi:10.1016/j.micpath.2006.11.003.

27. Lai SK, Wang YY, Wirtz D, Hanes J. Micro- and macrorheology ofmucus. Adv Drug Deliv Rev 61: 86–100, 2009. doi:10.1016/j.addr.2008.09.012.

28. Larsen TH, Furst EM. Microrheology of the liquid-solid transition duringgelation. Phys Rev Lett 100: 146001, 2008. doi:10.1103/PhysRevLett.100.146001.

29. Lysy M, Pillai NS, Hill DB, Forest MG, Mellnik JW, Vasquez PA,McKinley SA. Model comparison for single particle tracking in biologicalfluids (Abstract). J Am Stat Assoc 111: 1413–1426, 2017. doi:10.1080/01621459.2016.1158716.

30. Macauley MS, Crocker PR, Paulson JC. Siglec-mediated regulation ofimmune cell function in disease. Nat Rev Immunol 14: 653–666, 2014.doi:10.1038/nri3737.

31. Mason TG. Estimating the viscoelastic moduli of complex fluids using thegeneralized Stokes-Einstein equation. Rheol Acta 39: 371–378, 2000.doi:10.1007/s003970000094.

32. Matsui H, Grubb BR, Tarran R, Randell SH, Gatzy JT, Davis CW,Boucher RC. Evidence for periciliary liquid layer depletion, not abnormalion composition, in the pathogenesis of cystic fibrosis airways disease.Cell 95: 1005–1015, 1998. doi:10.1016/S0092-8674(00)81724-9.

33. Matsui H, Wagner VE, Hill DB, Schwab UE, Rogers TD, Button B,Taylor RM II, Superfine R, Rubinstein M, Iglewski BH, Boucher RC.A physical linkage between cystic fibrosis airway surface dehydration andPseudomonas aeruginosa biofilms. Proc Natl Acad Sci USA 103: 18131–18136, 2006. doi:10.1073/pnas.0606428103.

34. Matthews LW, Spector S, Lemm J, Potter JL. Studies on pulmonarysecretions. I. The over-all chemical composition of pulmonary secretionsfrom patients with cystic fibrosis, bronchiectasis, and laryngectomy. AmRev Respir Dis 88: 199–204, 1963.

35. Porterfield M, Zhao P, Han H, Cunningham J, Aoki K, Von Hoff DD,Demeure MJ, Pierce JM, Tiemeyer M, Wells L. Discrimination be-tween adenocarcinoma and normal pancreatic ductal fluid by proteomicand glycomic analysis. J Proteome Res 13: 395–407, 2014. doi:10.1021/pr400422g.

36. Radicioni G, Cao R, Carpenter J, Ford AA, Wang T, Li L, KesimerM. The innate immune properties of airway mucosal surfaces are regu-lated by dynamic interactions between mucins and interacting proteins: themucin interactome. Mucosal Immunol 9: 1442–1454, 2016. doi:10.1038/mi.2016.27.

37. Radtke T, Böni L, Bohnacker P, Fischer P, Benden C, Dressel H. Themany ways sputum flows. Dealing with high within-subject variability incystic fibrosis sputum rheology. Respir Physiol Neurobiol 254: 36–39,2018. doi:10.1016/j.resp.2018.04.006.

38. Ramsey K, Radicioni G, Hill D, Ehre C, Button B, Alexis NE, KnowlesM, Donaldson S, Kesimer M, McGuckin M, Boucher R. AirwaysMucus Pathogenesis in Patients with Non-Cystic Fibrosis Bronchiectasisand Primary Ciliary Dyskinesia. Dallas, TX: American Thoracic Society,2018, p. PA5049.

39. Raynal BDE, Hardingham TE, Thornton DJ, Sheehan JK. Concen-trated solutions of salivary MUC5B mucin do not replicate the gel-forming properties of saliva. Biochem J 362: 289 –296, 2002. doi:10.1042/bj3620289.

40. Round AN, Rigby NM, Garcia de la Torre A, Macierzanka A, MillsENC, Mackie AR. Lamellar structures of MUC2-rich mucin: a potentialrole in governing the barrier and lubricating functions of intestinal mucus.Biomacromolecules 13: 3253–3261, 2012. doi:10.1021/bm301024x.

41. Rubin BK, Ramirez O, Zayas JG, Finegan B, King M. Collection andanalysis of respiratory mucus from subjects without lung disease. Am RevRespir Dis 141: 1040–1043, 1990. doi:10.1164/ajrccm/141.4_Pt_1.1040.

42. Rubinstein M, Colby RH. Polymer Physics. Oxford, UK: Oxford Univ.Press, 2003.

43. Schnaar RL. Glycobiology simplified: diverse roles of glycan recog-nition in inflammation. J Leukoc Biol 99: 825–838, 2016. doi:10.1189/jlb.3RI0116-021R.

44. Schuster BS, Suk JS, Woodworth GF, Hanes J. Nanoparticle diffusionin respiratory mucus from humans without lung disease. Biomaterials 34:3439–3446, 2013. doi:10.1016/j.biomaterials.2013.01.064.

45. Thornton DJ, Rousseau K, McGuckin MA. Structure and function ofthe polymeric mucins in airways mucus. Annu Rev Physiol 70: 459–486,2008. doi:10.1146/annurev.physiol.70.113006.100702.

46. Varki A, Cummings RD, Aebi M, Packer NH, Seeberger PH, Esko JD,Stanley P, Hart G, Darvill A, Kinoshita T, Prestegard JJ, Schnaar RL,Freeze HH, Marth JD, Bertozzi CR, Etzler ME, Frank M, Vliegent-hart JFG, Lütteke T, Perez S, Bolton E, Rudd P, Paulson J, KanehisaM, Toukach P, Aoki-Kinoshita KF, Dell A, Narimatsu H, York W,Taniguchi N, Kornfeld S. Symbol nomenclature for graphical represen-tations of glycans. Glycobiology 25: 1323–1324, 2015. doi:10.1093/glycob/cwv091.

47. Wagner CE, Turner BS, Rubinstein M, McKinley GH, Ribbeck K.A rheological study of the association and dynamics of MUC5AC gels.Biomacromolecules 18: 3654 –3664, 2017. doi:10.1021/acs.biomac.7b00809.

48. Waigh TA. Advances in the microrheology of complex fluids. Rep ProgPhys 79: 074601, 2016. doi:10.1088/0034-4885/79/7/074601.

49. York WS, Agravat S, Aoki-Kinoshita KF, McBride R, Campbell MP,Costello CE, Dell A, Feizi T, Haslam SM, Karlsson N, Khoo KH,Kolarich D, Liu Y, Novotny M, Packer NH, Paulson JC, Rapp E,Ranzinger R, Rudd PM, Smith DF, Struwe WB, Tiemeyer M, Wells L,Zaia J, Kettner C. MIRAGE: the minimum information required for aglycomics experiment. Glycobiology 24: 402–406, 2014. doi:10.1093/glycob/cwu018.